# Models

Every model on the Jevals boards, scored on the same questions against human labels.

| Model | Type | noul Decision Score (rank) | choice Decision Score (rank) | score Decision Score (rank) |
| --- | --- | --- | --- | --- |
| [Gemini 3.8 Flash](https://jevals.com/models/gemini-3.8-flash/) | LLM, Google, reasoning low | 73.0 (tied for rank 1 of 8) | 74.1 (rank 1 of 8) | 4.6 (tied for rank 1 of 8) |
| [Jev](https://jevals.com/models/jev/) | System One, TypeSafe AI | 69.0 (tied for rank 1 of 8) | 67.8 (tied for rank 2 of 8) | 9.2 (tied for rank 1 of 8) |
| [GLM-5.3](https://jevals.com/models/glm-5.3/) | LLM, Z.ai, reasoning low | 60.6 (tied for rank 3 of 8) | 66.8 (tied for rank 2 of 8) | 7.8 (tied for rank 1 of 8) |
| [DeepSeek V4.1 Flash](https://jevals.com/models/deepseek-v4.1-flash/) | LLM, DeepSeek, no reasoning | 47.5 (rank 7 of 8) | 63.9 (tied for rank 4 of 8) | −19.0 (rank 7 of 8) |
| [Qwen3.8 Flash](https://jevals.com/models/qwen3.8-flash/) | LLM, Alibaba, no reasoning | 62.4 (rank 2 of 8) | 62.4 (tied for rank 4 of 8) | −1.4 (tied for rank 3 of 8) |
| [Mistral Medium 3.5](https://jevals.com/models/mistral-medium-3.5/) | LLM, Mistral AI, no reasoning | 58.0 (tied for rank 3 of 8) | 59.5 (rank 5 of 8) | −13.7 (rank 5 of 8) |
| [Mercury 2.5](https://jevals.com/models/mercury-2.5/) | LLM, Inception, no reasoning | 55.7 (tied for rank 3 of 8) | 54.0 (rank 7 of 8) | −5.5 (tied for rank 3 of 8) |
| [Label prior](https://jevals.com/models/label-prior/) | baseline | 0.0 (rank 8 of 8) | 0.0 (rank 8 of 8) | 0.0 (tied for rank 1 of 8) |

Decision Score: 100 = perfect, 0 = guessing the label base rates, below 0 = worse than that. Ranks: 1 + the number of rows significantly better (paired item bootstrap, 95%); rows that cannot be told apart share a rank. Hand-off at 95%: the share of decisions a model can take alone, at its own confidence threshold, while staying at least 95% right.

Source: https://jevals.com/models/ · release 2026-09-18, suite 0.1.0 · data (CC-BY-4.0): https://jevals.com/data/releases/2026-09-18/board.json · cite as: Jevals (jevals.com), release 2026-09-18, suite 0.1.0. CC-BY-4.0.